Dynamic Strongly Convex Sparse Operator with Learning Mechanism for Sparse Large-Scale Multi-Objective Optimization

Sparse large-scale multi-objective optimization problems (SLSMOPs) hold significant practical relevance across various domains. However, the efficacy of existing evolutionary algorithms (EAs) in tackling these optimization challenges is limited due to the high-dimensional search space and the sparsity inherent in Pareto optimal solutions. To overcome these difficulties, a dynamic strongly convex sparse operator with learning mechanism (DSCSOLM) is proposed. We design a novel strongly convex function that can effectively generate sparse solutions and enable the newly generated sparse solutions to learn knowledge from the Pareto optimal solutions, making the obtained sparse solutions more in line with the sparse distribution of the Pareto optimal solutions. Moreover, dynamic parameter is used within the proposed strongly convex function during the execution of the algorithm. Experimental results of benchmark and neural network training problems validate that DSCSOLM outperforms state-of-the-art (SOTA) comparative algorithms.